Adaptive-Sequential-Sampling

This code implements the adaptive sequential sample average approximation framework for solving two-stage stochastic programs. Several algorithms for solving individual SAA problems are implemented, including the adaptive partition-based level decomposition method.

Reference:

1. W. van Ackooij, W. de Oliveira, and Y. Song, “Adaptive Partition-based Level Decomposition Methods for Solving Two-stage Stochastic Programs with Fixed Recourse”, INFORMS Journal on Computing 30(1), 2018.
2. R. Pasupathy and Y. Song, "An Adaptive Sequential Sample Average Approximation Framework for Solving Two-stage Stochastic Programs", available on Optimization Online, 2019.

Requirements:

1. Commercial optimization software CPLEX
2. Open-source numerical computing package Eigen

Acknowledgement

This code was originally developed by Yongjia Song. It was then reorganized during Summer 2019 by Alicia Devasia and Casey Hird, two undergraduate students from Clemson University, who were supported by the Research Experience for Undergraduate program of the National Science Foundation.

How to Run CPLEX on Windows

Additional Requirements

Install Visual Studio 2019

• After installing, choose Desktop development with C++ workload

Create a Project

1. Once Visual Studio finished installing, click on Create a new project

2. Click on Empty Project with tabs C++, Windows, Console and create a Project Name

3. Now, add all the files above

   • Project >> Add Existing Item >> Select your files
     • Appropriate files will be sorted into the Header Files and Source Files

4. Next, you have to set some options so that the project knows where to find the CPLEX and Concert include files and CPLEX, Concert, and Eigen libraries.

   From the Project menu, choose "Project Name" Properties.The "Project Name" Property Pages dialog box appears.

   1. In the Configuration drop-down list, select Release.In the Platform drop-down list, select x64 to create a 64-bit application.

   2. Select C/C++ in the Configuration Properties tree.

      • Select General:
        • In the Additional Include Directories field, add the directories:
          • C:\Program Files\IBM\ILOG\CPLEX_Studio1290\cplex\include
          • C:\Program Files\IBM\ILOG\CPLEX_Studio1290\concert\include
          • Path to the eigen folder
            • E.g. C:\Users<Username>\downloads\eigen-eigen-323c052e1731
      • Select Preprocessor:
        • Add IL_STD to the Preprocessor Definitions field.
      • Select Code Generation:
        • Set Runtime Library to Multi-threaded DLL (/MD).

   3. Select Linker in the Configuration Properties tree.

      • Select General and then select **Additional Library Directories. Add the directories :
        • C:\Program Files\IBM\ILOG\CPLEX_Studio1290\cplex\include\lib\x64_windows_vs2015\stat_mda
        • C:\Program Files\IBM\ILOG\CPLEX_Studio1290\concert\lib\x64_windows_vs2015\stat_mda
      • Select Input and then select Additional Dependencies. Add the files:
        • cplex1290.lib
        • ilocplex.lib
        • concert.lib

   Click OK to close the "Project Name" Property Pages dialog box.

5. Next, you have to set the default project configuration. From the Build menu, select Configuration Manager…

   • Select Release in the Active Solution Configuration drop-down list.
     • Select x64 in the Active Solution Platforms drop-down list.
     • Click Close.

To Build/Compile the CPLEX Project

Finally, to build the project, from the Build menu, select Build Solution. After completion of the compiling and linking process, the target is created,"project name".exe.

To Run the CPLEX Project

Once a successful build is made, you are ready to run your program:

1. Tools >> Visual Studio Command Prompt
2. The program consists of 7 command line arguments:

   1. Path to executable file

      • The full path of the "project name".exe is typically the second to last line of the Output Box after a successful build.

   2. Path to instance (folder provided above)

   3. Path to a results folder

      • The output of the program is a text file that print the results of the chosen option by the user.

   4. Option:

      • -1 - extensive, 0 - Benders single, 1 - level, 2 - partly inexact bundle defined by partitions, 3 - sequential, 4 - adaptive, 5 - adaptive + partition, 6 - solve instances with a given # of samples in a retrospective way

   5. Suboption: only apply for option = 3, 4, 5

   6. Alternative use of suboption for option = 6: # of samples used

   7. Random number

E.g. of command line arguments

"C:\Users<username>\source\repos<project name>\x64\Release<project name.exe" "C:\Users<username>\source\repos<project name>\instances\20x20-1-20000-1-clean.dat" "C:\Users<username>\source\repos<project name>\results\temp" 0 0 0.5 1e-3 5